Should we categorize triangulated groups in GEDmatch?

Question

GEDmatch (and 23andMe) reveal Trangulated Groups (TGs) for testers with Family Finder, 23andMe or AncestryDNA.   Each TG is composed of actual cousins who share the same ancestor (or ancestral couple) and in many cases are each motivated research groups.

Can we propose a standard syntax to categorize testers who belong to the same TG?

Something like TG12345C11B11.4E16.1 ?  Which translates as Trangulated Group number 12345 Chromosome 11 Beginning 11.4 million Ending 16.1 million.  Can we think of something better?

Comments

What is the reason to categorize them?

---

Each Triangulated Group (TG) consists of cousins (who all match each other on the same segment) with a shared ancestor.  Categorizing TGs would help us keep track of them.

For example, Triangulated Group "TG12345C11B11.4E16.1" consists of Doe-123, Smith-456, and Jones-321 who don't yet have a paper connection but each have an ancestral branch from the same hollow/holler in Kentucky.  It would make it easier to do a "what if".  What if Doe-123's ancestor Jane Baker and Jones-321's ancestor Sarah Pennybaker where sisters? Can we find direct maternal line descendants of each to see if they are an mtDNA match?

---

It will help you for sure (with keeping track). But I thought categories are for others to find something relevant. Those that are in the TG should know about it through other means of communication (emails, forums etc.).

But maybe it's my (mis-) understanding of categories here at WT.

---

That's very simular to the idea I've been forming in the dark places of my mind. [grin] - Tags that represented halpogroups, but it seems that it might be hard to define.

---

Unfortunately you can only have a total of 20 tags (and I've already reached my limit)-:

Answer 1

An ancestral couple can have descendants that form several triangulated groups. One kit could belong to more than one of those groups. Concentrating on just one segment may not be the way to go.

Answer 2

That's a tough one, Peter. I thought about that one for months for my new website but I couldn't come up with a satisfying solution.

Also James is correct, different parts of the DNA of the Common Ancestor (CA) will form Triangulated Groups (TG's) on different chromosomes. If all lines were continued until current days, basically some 200+ TG's could exist from the same CA.

Also, let's keep in mind that even though the DNA is only from one of the Ancestor couple (either father or mother), we can't distinguish with current testing methods who gave the DNA to the TG.

Therefore the TG is always a couple (until the testing companies can produce phased results) even though one of them didn't gave any DNA to that ancestral segment which everyone in the TG shares.

Answer 3

I agree that this would be a great idea, but the number of potential groups is probably too large to be useful I would think  AND it would be hard to educate everyone how to use it properly.

It is similar to my AncestryDNA Circles idea - http://www.wikitree.com/g2g/205350/category-for-ancestrydna-circles?show=205350#q205350 but at least under this approach the category is already linked to the ancestor, not testers to an unknown ancestor.

I suspect it could work for personal categories.  I'm flagging 'ancestors of interest' in a category I set up for my autosomal tests.  These are linked to my Coat-12 ID but have also linked them up to a higher level category for Autosomal DNA confirmations.  http://www.wikitree.com/wiki/Category:Autosomal_DNA_ConfirmationsI see a couple of others have put information there too

Comments

Hello Veronica,

I do believe it would be too hard to educate everyone how to use it properly...

Instead of categories, I wonder if there could be input fields for variables (chromosome #, start, stop, cM, WikiTree IDs of those in the TG,, etc) and each record entry generated a segment on a chromosome browser like that at http://tinyurl.com/z975k9y

Clicking on a segment would reveal the WikiTree IDs for that TG and its other field details.

Sincerely, Peter

---

That sounds like a feasible idea.  But wouldn't there still be thousands of potential groups?  Also, for most of my TG's I might be the only one with a wiki-id.  GED ID's would be more available, you probably need to limit whatever the program does to at least having a GedID.. Another field is probably Wikitree contact or 'added by'.

Perhaps a mix of categories and free space pages might work. The category could be 'Autosomal DNA Triangulated Groups' , then users could each have a free space page, called maybe ''Autosomal DNA Triangulated Groups - WIKI-ID', then within that you could have these tables maybe?

But I think you are trying to find a way to pool it all together to be useful to others.  Maybe free space pages by chromosome number?  Then there could be a table where data was added?  

If you could put it all in one free space page, using the method you originally suggested, could the input fields ensure there are at least 3 persons to a group before it accepted the data?  How would we ensure that groups being entered all matched each other?  Could the process/output be able to work out whether there were multiple TG's or would it just be compiling all the data into one big sheet?

I think the idea has merit.  Having one big table is probably more useful for others.  Although perhaps if it started off as an idea for use with personal categories a few individuals could trial it, see how it goes and work out how it might work best for a whole of Wikitree thing?

---

Hello Veronica,

I imagine only caring about TGs for me and my relatives (but not their TGs due to their ancestry that I don't share).  So I don't see it as too large a number of groups to manage.  So each auDNA tester could do the same.

These TGs already exist in GEDmatch's triangulation report, but it is difficult to quickly find their shared ancestry.  This is possible with WikiTree's Relative Finder.  Which is why it would be good to have an elegant utility which associates TGs with shared ancestry.

Sincerely, Peter

---

Well in that case,

1.  Add a category into the current structure:

Autosomal DNA (already exists)

>  Autosomal Triangulated Groups  (new)

2.  Then create a Free Space page for each Wikitree user who wants one with a consistent naming convention-

* Autosomal Triangulated Groups - Wiki-ID

3.  Link the Free Space Profile page to both of these categories:-

*  Autosomal Triangulated Groups

* Wiki-ID (personal category)

4. Add the data to the free space page as suggested by you.

I am comfortable with 1-3.  

Thinking about how best to display the groups (4) still remains a challenge. Do we want to have a go and see what it looks like.  I have some TG data I could start with.  If you could build the tool I could be a volunteer to see how it works for me?

Could we somehow have a field for possible ancestors on wikitree, although maybe that would be best for notes on the free space page under each group?  Just thinking about my 'ancestor of interest' idea.  At the moment I go back to my category page and see if one is listed already when I get another match.  I am building up a picture but free space page is probably better than my current cut and pastes to Evernote!

---

Now we get into my stumbling blocks and territory I know little about ;-)

---

I just ran triangulated results and found this thread trying to figure out if there was somewhere to put them.  I'm happy to share if someone has a template or sample to follow.

---

Unless the testers are on WIkitree I find it hard to envisage how it could work other than by using a free space page and linking it to categories as I've already suggested.

Whilst my original thought was by wiki_id perhaps a better way would be by gedmatch ID.    With the gedmatch ID linked to the wiki-id of the tester, that way the list would be available by GedID and could be browsed by others.

So maybe, 

1.  Add categories into the current structure:
* Autosomal DNA (already exists)
>  Autosomal Triangulated Groups by Gedmatch ID (new)

> Autosomal Triangulated Groups by Wiki-ID (new)

2.  Then create a Free Space page for each Wikitree tester who wants one with a consistent naming convention and linked to wiki id's, which allows for multiple testers managed by one wiki-id.
* Autosomal Triangulated Groups - Gedmatch ID i.e. FXXXXXX 

Link up to to WIKItree member

>Autosomal Triangulated Groups - Wiki-ID i.e. COAT-12 (create new)

3.  Link the Free Space Profile page to the category
*  Autosomal Triangulated Groups by Gedmatch-ID

4. Link the Autosomal Triangulated Groups - Wiki-ID to the higher level category and your personal category, i.e.
* Wiki-ID (personal category), e.g. Coat-12

* Autosomal Triangulated Groups by Wiki-ID

5. Add the data to the free space page as suggested by Peter.  How we do this bit needs more work.

I was going to spend some time trying to work on an example but just haven't had time of late.  If any one has any ideas it would be great to hear them.  I think like Peter, by chromosome, probably just using the data the same way it is produced by the Tier 1 gedmatch triangulation report would be the easiest.

---

I think it would be useful for mapping to distinguish between different scenarios of interest.

If I am mapping and documenting a relationship between myself and someone else on WikiTree then it suffices to have a template or similar that allows me to note the relationship by what portion of what chromosome I share with them with what tested tolerances. I think this forms the basis of the triangulation grouping problem as all triangulation grouping solutions will necessarily depend on how equivalent pairing is handled. I would like to see this kind of functionality as the output of the relationship finder, or I would like to be able to input my testing results into the relationship finder; most of the people that I flagged for genetic comparison based on our common genealogical ancestry turned out to be negatives.

Maybe a field for entering genetic relations of known or unknown type. Similar to siblings on the profile. Triangulation groups could come off of that arrangement.

 

In a bigger picture, each profile can be represented by 23 or more chromosome pairs plus mitochondrial DNA, and each person's genetic genealogy can be represented by a chromosome map with haplogroups. The map can generally be bifurcated by paternal and maternal sides; at current, it would be impractical to sort all the chromosomes alleles, but it isn't technological infeasible in the short term, and in some people's cases, they'd be able to distinguish which parts of which pairs they inherited from whom out to a couple of generations with comparisons to close relatives. This wouldn't necessarily be immediately useful for distinguishing most common recent ancestors, but it could be used heuristically to figure out which branches of the family graph have to grow together but haven't been genealogically documented yet or have been but likely have issues with their status.